Researchers tested the impact of contextual mismatch, proactive interference, and working memory (WM) on toddlers’ transfer across contexts. Forty‐two toddlers (27–34 months) completed four object‐retrieval trials, requiring memory updating on Trials 2–4. Participants watched hiding events on a tablet computer. Search performance was tested using another tablet (match) or a felt board (mismatch). WM was assessed. On earlier search trials, WM predicted transfer in both conditions, and toddlers in the match condition outperformed those in the mismatch condition; however, the benefit of contextual match and WM decreased over trials. Contextual match apparently increased proactive interference on later trials. Findings are interpreted within existing accounts of the transfer deficit, and a combined account is proposed.